Create a Scale Group
Use this guide to create an Auto Scaling scale group from the MyAccount portal. A scale group launches and terminates compute nodes automatically against the policy you define, using a saved image and a compute plan you select.
If you are new to Auto Scaling, read Auto Scaling Concepts first. You can also create and manage scale groups through the Auto Scaling API.
Before You Begin
- Prepare a saved image. Configure a node so your application starts at boot, save an image from it, and wait until the image reaches the Ready state.
- Decide your plan, node counts, and policy. Know the vCPU/RAM/storage you need, your minimum/maximum node counts, and whether you will scale on CPU, a custom metric, a schedule, or a combination.
- Plan to attach a load balancer. A scale group is most useful behind a load balancer, which keeps a stable address as nodes change.
Scale groups launch billable compute nodes. You are charged hourly for every running group node. Review your minimum and maximum node counts before you create the group.
Step 1: Select a Saved Image
- Log in to MyAccount.
- In the left navigation, go to Compute > Auto Scaling.
- Start the create flow (Get Started on the empty state, or Add New Node / the create button when groups already exist).
- On Select Saved Image, the table lists each image's Saved Image Name, Creation Date, and Status. Only images in the Ready state are listed.
- Select an image, then continue.
You cannot delete a saved image while a scale group is using it. Terminate the scale group first.
Step 2: Select a Plan
Choose where the group nodes run and the hardware they use.
- Choose the cluster type — E2E Cluster (shared) or Private Cluster (dedicated hosts). For a private cluster you enter the cluster name and the CPU, RAM, and storage for each node instead of picking a public plan.
- For an E2E cluster, filter the plan list with the All vCPU, ALL RAM, and All Storage dropdowns (use Reset to clear filters).
- Review each plan's Family, OS, Plan, vCPU, IOPS, RAM, Storage, and Price (per month and per hour).
- Select a plan. If a plan shows Request instead of Select, it is not currently available in that location.
Plan capacity and pricing change over time. Use the live plan list in the portal as the source of truth.
Step 3: Configure Scale Group Settings
This step defines the group's name, node counts, policy, network, and security.
Scale Group Name
Enter a Scale Group Name. It must be 2 to 40 characters and may contain letters, digits, underscores, and hyphens only.
Scale Group Parameters
| Field | What to enter |
|---|---|
| Minimum Node | The floor for the pool. Must be at least 1 and less than the maximum (and less than the downscale cardinality). Default 2. |
| Maximum Node | The ceiling for the pool. Must be greater than the minimum (and greater than the upscale cardinality) and cannot exceed 50. Default 5. |
On a private cluster, autoscaling depends on host availability at the scheduled time.
Select a Policy
Choose how the Scaler decides when to add or remove nodes:
- Elastic Policy — auto-adjusts capacity based on the utilization of a chosen parameter.
- Scheduled Policy — adjusts capacity at specific times using cron expressions.
- Elastic and Scheduled Policy — both together. This is the default.
The matching policy panels appear below based on your choice.
Elastic Policy
The elastic policy scales the group when a monitored metric crosses your threshold for a sustained watch period.
Policy Parameter Type — choose one:
- Default — select the built-in parameter. The available default parameter is CPU.
- Custom — enter a Policy parameter name (for example,
NETWORK_TRAFFIC). The name must be uppercase letters, digits, and underscores, must start and end with a letter or digit, and cannot be a reserved default name. Your application is responsible for publishing this metric to the node — see Custom Scaling Policies.
| Field | What it controls |
|---|---|
| Add node when … | The scale-up threshold and operator — more than (>) or more than equal to (>=). For the default (CPU) parameter the value must be greater than 30 and less than 100. Default 60, operator >. |
| Remove node when … | The scale-down threshold and operator — less than (<) or less than equal to (<=). Must be less than half the scale-up value; for the default parameter it must be greater than 10 (greater than 0 for a custom parameter). Default 30, operator <. |
| Watch Period | Number of consecutive periods the condition must hold. Default 3. |
| Period duration(s) | Length of each period, in seconds. Default 10. |
| Cooldown(s) | Pause after a scaling action, in seconds. Default 150. |
The panel shows a plain-language summary, for example: "The scale group will watch for 3 consecutive periods (of 10 seconds each) when CPU utilization stays above 60% or falls below 30%. When such a condition occurs the scaling operation will be initiated."
You only define the upscale and downscale rules — the Scaler creates the underlying negative policy automatically.
Scheduled Policy
The scheduled policy changes capacity at fixed times, which is ideal for predictable traffic (for example, busier on weekday mornings). It has two recurrences:
Upscale Recurrence
| Field | What to enter |
|---|---|
| Recurrence | A cron expression (for example, 0 9 * * 1-5 = 9:00 AM, Monday–Friday). Pick from the suggestions or type your own. If valid, the portal shows the scheduled time in plain language. |
| Cardinality | The total number of running nodes after the cron fires. Must be ≥ the downscale cardinality and ≤ the maximum nodes. Default 4. |
Downscale Recurrence
| Field | What to enter |
|---|---|
| Recurrence | A cron expression (for example, 0 0 * * 6,7 = 12:00 AM, Saturday and Sunday). Must differ from the upscale recurrence. |
| Cardinality | The total number of running nodes after the cron fires. Must be ≤ the upscale cardinality and ≥ the minimum nodes. Default 2. |
The Cron expression 0 0 * * * runs at midnight every day. Set the upscale cardinality below your maximum nodes, and the downscale cardinality above your minimum nodes.
Network and Public IP
Open the VPC panel to attach the scale group to a VPC. On an E2E cluster a VPC is optional; on a private cluster it is required. Select a VPC from the dropdown (only VPCs in the Active state can be selected), or use the refresh and Create New VPC controls.
When a VPC is selected, a checkbox appears: "Enable this option to assign static public IP to scale group." Enable it if group nodes need a public IP.
Security (Encryption)
Open the Security panel to encrypt group nodes at rest. Select Enable Encryption, then optionally enter a passphrase. See Auto Scale Encryption for the passphrase rules and full details.
Encryption can be enabled only at creation time and cannot be turned off afterwards.
Security Group
Open the Security Group panel and choose an existing security group, or select Create New Security Group. A scale group must always have at least one security group attached. If you have none, create one and select it.
Step 4: Create the Scale Group
Review the Summary panel, then select Create Scale.
The new group begins in the Creating state (shown while the underlying service deploys). When it reaches Running, the Scaler starts the desired number of nodes and the details tabs become available.
After creation you can monitor and operate the group from its details page — see Manage Scale Groups.
Create an Encrypted Scale Group
The flow is identical to the above, with one addition at Step 3 → Security: select Enable Encryption before creating the group, and optionally set a passphrase. Because encryption is creation-time only, you cannot add it to an existing group later. For the full walkthrough, see Auto Scale Encryption.
Related Resources
| Resource | Use it for |
|---|---|
| Auto Scaling Concepts | Terminology and how policies work. |
| Custom Scaling Policies | Publish and scale on your own metric. |
| Auto Scale Encryption | Encrypt group nodes at rest. |
| Manage Scale Groups | Tabs, actions, and editing after creation. |
| Backend Mapping | Attach the scale group to a load balancer. |
| Create a Scale Group (API) | Automate creation. |